Nondynamic and Dynamic Kinetic Resolution of Lactones with

This key intermediate, expected to be configurationally unstable,6 should best22 be synthesized via an intramolecular biaryl coupling of the carboxyli...
2 downloads 0 Views 135KB Size
Download PDF
J. Am. Chem. Soc. 2000, 122, 9127-9133

9127

Nondynamic and Dynamic Kinetic Resolution of Lactones with Stereogenic Centers and Axes: Stereoselective Total Synthesis of Herbertenediol and Mastigophorenes A and B† Gerhard Bringmann,*,‡ Thomas Pabst,‡ Petra Henschel,‡ Ju1 rgen Kraus,‡ Karl Peters,§ Eva-Maria Peters,§ David S. Rycroft,| and Joseph D. Connolly| Contribution from the Institut fu¨ r Organische Chemie, UniVersita¨ t Wu¨ rzburg, Am Hubland, D-97074 Wu¨ rzburg, Germany, Max-Planck-Institut fu¨ r Festko¨ rperforschung, Heisenbergstrasse 1, D-70506 Stuttgart, Germany, and Department of Chemistry, UniVersity of Glasgow G12 8QQ, Scotland, U.K. ReceiVed April 26, 2000

Abstract: The stereoselective total synthesis of the sesquiterpene herbertenediol (3) and of its naturally occurring dimers, mastigophorenes A [(P)-1] and B [(M)-1], is described. Following the “lactone concept”, the configuration at the biaryl axis was atropo-divergently induced to be P or, optionally, M, by stereocontrolled reductive ring cleavage (diastereomeric ratio up to 97:3) of the configurationally unstable joint biaryl lactone precursor 17 using the oxazaborolidine-borane system, through dynamic kinetic resolution. Mechanistic considerations of the lactone coupling suggested interference by a methoxy group next to the halogen substituent and led to an improvement of the coupling yield from 39 to 87% (to give the lactone 37). As a new, likewise highly efficient variant of the lactone method, we report for the first time thesnow nondynamicskinetic resolution of a structurally related, but centrochiral “aliphatic-aromatic” lactone, (rac)-10. Its highly efficient (krel > 300) enantiomer-differentiating Corey-Bakshi-Shibata reduction delivers the centrochiral building block (R,R)-10 in good chemical yield and with excellent stereochemical purity (enantiomeric excess > 99.9%; enrichment of the starting matrial). The new synthesis of natural herbertenediol (3) confirms its absolute stereostructure as well as that of its dimers, (P)-1 and (M)-1.

Introduction The natural products mastigophorene A [(P)-1] and B [(M)1] are representatives of a group of axially chiral “dimeric”1 sesquiterpenes exhibiting nerve growth stimulating activity.2 Such nonpeptidyl neurotrophic substances are regarded as promising potential therapeutic agents for degenerative diseases of the central nervous system such as Parkinson and Alzheimer.3 Therefore the total synthesis of (P)-1 and (M)-1 is an attractive target, even more so because the interesting, C2-symmetric structures involve elements of both axial and centro chirality. In our investigations, we published two different strategies for the construction of the simplified analog 2,4 the first one by a (nonstereoselective) biomimetic oxidative phenolic coupling, leading to the likewise neurotrophic5 mastigophorene analog 2,4 and the second one using our “lactone methodology”6 for † Part 87 of the series “Novel Concepts in Directed Biaryl Synthesis”. For part 86 see ref 18. * To whom correspondence should be addressed (Tel: +49 931 888 5323; Fax: +49 931 888 4755; E-mail: [email protected]). ‡ Universita ¨ t Wu¨rzburg. § Max-Planck-Institut, Stuttgart. | University of Glasgow. (1) Strictly speaking such “dimers” should be described as “dehydrodimers”. (2) Fukuyama, Y.; Asakawa, Y. J. Chem. Soc., Perkin Trans. 1 1991, 2737. (3) Skaper, S. D.; Walsh, F. S. Mol. Cell. Neurosci. 1998, 12, 179. (4) Bringmann, G.; Pabst, T.; Busemann, S.; Peters, K.; Peters, E.-M. Tetrahedron 1998, 54, 1425. (5) Gille, G.; Pabst, T.; Bringmann, G.; Janetzky, B.; Reichmann, H.; Rausch, W.-D. Drug DeV. Res. 2000, 50, 153. (6) Bringmann, G.; Breuning, M.; Tasler, S. Synthesis 1999, 525.

the atropo-enantioselective construction of the biaryl axis. An application of the biomimetic strategy to a simple derivative of the authentic monomeric half, 3, led to the first formal total synthesis of (P)-1 and (M)-1, albeit with moderate asymmetric inductions.7 During our ongoing work, Meyers et al. published a synthetic pathway to (P)-1 and (M)-1, involving a first enantioselective route to (-)-herbertenediol (3) and an atropodiastereoselective Ullmann coupling of chiral aryloxazolines.8 We report on our highly stereoselective total synthesis of mastigophorenes A [(P)-1] and B [(M)-1] by a 2-fold application of the lactone method:6 first, the novel nondynamic kinetic resolution of “aromatic-aliphatic” lactones for the enantioselective synthesis of the centrochiral molecular “half”, herbertenediol (3), and second, the dynamic kinetic resolution of a configurationally unstable biaryl lactone for the directed, atropo-diastereodivergent construction of the biaryl array to give, optionally, the M- or the P-configured product, in high atropodiastereomeric ratios. (7) Bringmann, G.; Pabst, T.; Rycroft, D. S.; Connolly, J. D. Tetrahedron Lett. 1999, 40, 483. (8) Degnan, A. P.; Meyers, A. I. J. Am. Chem. Soc. 1999, 121, 2762. (9) Fukuyama, Y.; Kiriyama, Y.; Kodama, M. Tetrahedron Lett. 1996, 37, 1261. (10) Weeratunga, G.; Jaworska-Sobiesiak, A.; Horne, S.; Rodrigo, R. Can. J. Chem. 1987, 65, 2019. (11) Shiba, T.; Cahnmann, H. J. J. Org. Chem. 1964, 29, 3061. (12) Bringmann, G.; Schneider, S. Synthesis 1983, 139. (13) Kla¨rner, F.-G.; Adamsky, F. Chem. Ber. 1983, 116, 299. The elimination step to ethyl 1,2-dimethyl-cyclopent-2-enoate was done by azeotropic removal of water with toluene and p-toluenesulfonic acid. Subsequent hydrolysis as described gave (rac)-8 in 72% yield (2 steps).

10.1021/ja001455r CCC: $19.00 © 2000 American Chemical Society Published on Web 09/07/2000

9128 J. Am. Chem. Soc., Vol. 122, No. 38, 2000

Bringmann et al. Scheme 1

Scheme 2

Results and Discussion Our concept for the enantioselective synthesis of herbertenediol (3) was related to a synthetic pathway previously developeds albeit in a racemic formsby Fukuyama et al.9 (Scheme 1). In that approach the chiral quaternary carbon center had been built up by a diastereoselective intramolecular Heck coupling of (rac)-4 to give lactone (rac)-5, which was then converted to (rac)-3, including introduction of the second phenolic oxygen function. For the directed synthesis of enantiomerically pure 3, with its two oxygen functions, we modified this synthetic concept in two respects. First, to avoid the subsequent introduction of the second phenolic function, we started directly from the dioxygenated aromatic building block 7 (Scheme 2), which might be prepared either according to a method described in the literature10 or, more easily, from the known11 5-iodovanilline (6), by reaction with NaBH4 and reductive deoxygenation via the corresponding bromide.4,12 Esterification of 7 with the known13 acid (rac)-8 using dicyclohexylcarbodiimide (DCC) as the condensation agent gave (rac)-9. This iodo ester was, similar to (rac)-4 in the literature, diastereoselectively cyclized by a palladium-catalyzed Heck reaction, to afford, after hydrogenation, the cis-configured lactone (rac)-10 in racemic form (of which Scheme 2 arbitrarily shows the R,R-enantiomer). The cis-configuration, which could not be assigned by nuclear Overhauser effect (NOE) experiments, was clearly established by X-ray structure analysis of (rac)-10 (see Scheme 2). This analysis further showed the methyl group β to the carbonyl function to be in an axial position, and the R-methyl group in an equatorial position. This conformation avoids 1,3-allylic strain between the β-CH3 group and the aromatic hydrogen. The second, even more significant improvement of the synthetic pathway to 10 was development of a route to enantiomerically pure material. The acid 8, however, which would thus be required in stereochemically homogeneous form, had been synthesized only in an insufficient enantiomeric purity (enantiomeric excess (ee) ) 90%) in the literature.9 Therefore, given our good results with atropisomer-selective cleavage reactions of biaryl lactones, a kinetic resolution at the level of the aromatic-aliphatic lactone (rac)-10 seemed to offer an ideal extension of our lactone method. For this purpose, the rigid structure of 10 might indeed provide a good precondition for efficient stereodifferentiation of the two enantiomers. Given the mostly excellent results with oxazaborolidine-mediated borane reductions (Corey-Bakshi-Shibata (CBS) reaction)14 in the kinetic resolution of configurationally unstable6sor stable15s

Scheme 3

biaryl lactones, we then tried this method also for 10, i.e., for the first time in the enantiomer-differentiating reduction of nonbiarylic chiral lactones. Initial reduction experiments with (rac)-10 on an analytical scale, even when an excess of 11 was used, revealed the system to be too unreactive to be reduced completely to the diol 14 and the reaction stopped at the lactol 12 (obtained as a diastereomeric mixture) (Scheme 3). Fortunately, this reduction was found to proceed with perfect enantiomer-differentiating selectivity: After 51.5% conversion,16 the starting material 10 was enriched to an ee > 99%; the other lactone enantiomer was no longer detectable by HPLC analysis on a Chiralcel OD-H phase. Attempts to run the reaction with (14) Corey, E. J.; Helal, C. J. Angew. Chem., Int. Ed. Engl. 1998, 37, 1986. (15) Bringmann, G.; Hinrichs, J. Tetrahedron: Asymmetry 1997, 8, 4121. (16) The conversion and the relative rate krel were calculated according to (a) Kagan, H. B.; Fiaud, J. C. Top. Stereochem. 1988, 18, 249. (b) Chen, C.-S.; Fujimoto, Y.; Girdaukas, G.; Sih, C. J. J. Am. Chem. Soc. 1982, 104, 7294.

Kinetic Resolution of Lactones

J. Am. Chem. Soc., Vol. 122, No. 38, 2000 9129

Figure 2. Comparison of the experimental CD spectrum of 10 with the spectra calculated for its two possible enantiomeric forms.

Scheme 4

Figure 1. Rationalization of the observed stereodifferentiation and prediction of the stereoisomer expected to be reduced.

catalytic amounts of the oxazaborolidine 11, however, failed: With 0.1 equivalents of 11, only 8.4% conversion was observed. Nevertheless, to perform the reaction as economically as possible with respect to the chiral reagent, we used 0.6 equivalents of 11 in preparative reactions and then isolated enantiopure 10 (ee > 99.9%, krel > 300!)16 in an excellent 48% chemical yield () 96% based on the maximum possible yield for that enantiomer). For a directed synthesis of the correct enantiomer of natural herbertenediol (3), independent of a “late” assignment of the product configuration at the level of the stereochemically known17 monomeric target molecule 3, the absolute product configuration was already determined for the enantiopure lactone 10. From model considerations (Figure 1) based on previous experimental and computational investigations14,18,19 but also on the crystal structure of 10 (see above), a distinctly preferred attack of the N-activated borane to the B-activated carbonyl function of 10 should occur from a conformation related to (R,R)- or, (S,S)-13a [) (S)-11‚BH3‚10], thus avoiding a strong heteroallylic strain as, for example, in (R,R)-13b. Of the two isomers of 13a, (S,S)-13a should react much more easily, because here (assuming a conformation of the lactone part similar to that of 10 in the crystal) the quasi-intramolecular hydride attack would occur syn to the equatorial methyl group, but anti to the propylene bridge, which is axial in the R-position next to the carbonyl function. For (R,R)-13a, by contrast, the attack would have to occur syn to that axial part of the propylene bridge, so that this isomersand thus (R,R)-10 - should remain unreacted with (S)-11‚BH3. This assumption was confirmed by quantum chemical circular dichroism calculations.20 Thus, the circular dichroism (CD) spectrum calculated for (R,R)-10 shows good agreement with the experimental spectrum of the product 10 that remained unreacted with (S)-11, whereas the spectrum calculated for (S,S)10 is virtually opposite to the experimental spectrum (Figure 2), clearly indicating that the product (R,R)-10 needed for the synthesis of natural i.e., S-configured, herbertenediol (3), was (17) Matsuo, A.; Yuki, S.; Nakayama, M. J. Chem. Soc., Perkin Trans. 1 1986, 701. (18) Bringmann, G.; Hinrichs, J.; Kraus, J.; Wuzik, A.; Schulz, T. J. Org. Chem. 2000, 65, 2517. (19) Bringmann, G.; Vitt, D. J. Org. Chem. 1995, 60, 7674. (20) Bringmann, G.; Busemann, S. In Natural Product Analysis; Schreier, P., Herderich, M., Humpf, H.-U., Schwab, W., Eds.; Vieweg: Braunschweig, 1998; pp 195-212.

Scheme 5

that obtained with (S)-11. Conversely, with the ultimate outcome of the synthesis (see below), the CD calculations confirm the absolute centrochiral S-configuration for both herbertenediol (3) and the mastigophorenes A [(P)-1] and B [(M)-1] as established earlier.2,17 Starting from (R,R)-10, enantiomerically pure herbertenediol (3) was synthesized by analogy to the procedure described for racemic material by Fukuyama et al.9 Reduction with LiAlH4 (LAH) gave the diol 14 (see Scheme 4), which was converted to the still O,O-dimethylated herbertenediol precursor 15 by O-protection with Me2SO4, Swern oxidation, and WolffKishner reduction. Deprotection with BBr3 led to herbertenediol (3), which was found to be enantiopure by HPLC analysis on a Chiralcel OD-H phase, and was fully identical to natural material by its physical, chiroptical, and spectroscopic data.17 Another approach to enantiomerically pure 3, recently described in preliminary form,7 is its partial synthesis from the natural aldehyde 16 (Scheme 5), which co-occurs with herbertenediol (3) in the liverwort Herbertus aduncus and was isolated as described previously.21 Protection of the phenolic functions of 16 with Me2SO4 and reduction of the aldehyde function with LAH led to the corresponding alcohol, whose reductive deoxygenation via the corresponding bromide12 gave the dimethyl ether 15, which was deprotected with BBr3 to afford (-)-herbertenediol (3). This partial synthesis of 3 simultaneously confirms the S-configuration of 16, which had so far been assumed only from its slightly negative RD.21 (21) Buchanan, M. S.; Connolly, J. D.; Rycroft, D. S. Phytochemistry 1996, 43, 1245.

9130 J. Am. Chem. Soc., Vol. 122, No. 38, 2000

Bringmann et al.

Scheme 6

With enantiopure herbertenediol (3) in hand by partial and by total synthesis, we could now start building its natural dimers, mastigophorenes A [(P)-1] and B [(M)-1], based on previous experience from our model studies.4 Using our lactone method,6 both (P)-1 and (M)-1, optionally, should be synthesizable from the same biaryl lactone 17 (Scheme 6), by atropo-diastereoselective ring cleavage. This key intermediate, expected to be configurationally unstable,6 should best22 be synthesized via an intramolecular biaryl coupling of the carboxylic ester 18, and thus from the phenol 19 and the acid 20, both ultimately arising from herbertenediol (3). The acid 20 was synthesized from herbertenediol dimethyl ether (15) by bromination and cobalt-catalyzed oxidation with O2 (Scheme 7). Alternatively, a second approach to 20sby partial synthesisssucceeded by bromination of the natural21 aldehyde 16, protection with Me2SO4, and oxidation with NaClO2. The phenolic precursor 19 was obtained7 from 3 by selective benzylation of the less hindered oxygen at C-1 and subsequent methylation of the second phenolic function at C-2. Hydrogenolysis of the benzyl group then generated the monophenolic building block 19. From the phenol 19 (via its phenolate) and the acid 20 (via its acid chloride) the bromo ester 18 was obtained and submitted to a Pd0-catalyzed intramolecular biaryl coupling reaction, to give the desired key biaryl lactone 17, in 39% yield (subsequently improved, see later), together with 41% of recovered substrate. Longer reaction times did not lead to higher yields, but rather to decomposition of the product already formed. As expected, 17 consists of two rapidly interconverting atropodiastereomers (P)-17 h (M)-17 in solution, as required for a dynamic kinetic resolution in the ring cleavage reaction. In the crystal however, it occurs as theslike-configuredsP-atropisomer, exclusively, as shown by an X-ray structure analysis (see Scheme 7). In accordance with the predominance of (P)-17 in the crystal is the observation that LAH reduction of 17 gave the corresponding diols (P)-21 and (M)-21 in favor of the P-configured product (diastereomeric ratio (dr) 77:23). Rapid reduction by addition of crystalline 17 to a 1 M solution of LAH in tetrahydrofuran (THF) (-100 f -30 °C) gave an even better dr of 81:19. (22) For the synthesis of relatedsalbeit sterically less hinderedsbiaryl lactones by other methods, see: (a) Wang, W.; Snieckus, V. J. Org. Chem. 1992, 57, 424. (b) Davies, D. I.; Waring, C. J. Chem. Soc. C 1967, 1639 and references cited therein. (c) de Frutos, O.; Echavarren, A. M. Tetrahedron Lett. 1996, 37, 8953.

Still better diastereoselectivities and the option of producing either of the two atropisomers, (P)- or (M)-21, highly selectively from the same configurationally unstable lactone precursor 17, within a now dynamic kinetic resolution, were attained using Scheme 7

Kinetic Resolution of Lactones Scheme 8

techniques developed previously within the lactone methodology.6 Thus, reaction of (P)-17 h (M)-17 with (S)-11 gave (P)21 with an excellent dr of 97:3 (Scheme 8). But even the M-configured atropisomer of 21 (less readily produced using achiral reagents, see above) could be obtained with a good dr of 92:8 by using (R)-11, thus overcoming the molecule-inherent P-preference by a strong external asymmetric induction exerted by the CBS system. In both cases diastereomerically pure compounds were obtained by column chromatography. Further transformation of (P)- and (M)-21 to mastigophorenes A [(P)-1] and B [(M)-1], respectively (Scheme 8), followed protocols developed previously for related model compounds.4 Protection of the phenolic function of (P)- and (M)-21 by O-methylation, and reductive benzylic deoxygenation12 (via the corresponding bromide) provided the tetra-O-methylmastigophorenes (P)- and (M)-22, respectively. Deprotection with BBr3 gave (P)-1 and (M)-1, which proved to be identical in all respects with the data published2 for the natural products. Within this stereoselective synthesis of the mastigophorenes, their axial configuration was proved rigorously by an X-ray

Figure 3. X-ray crystal structure of tetra-O-methylmastigophorene A [(P)-22].

J. Am. Chem. Soc., Vol. 122, No. 38, 2000 9131 Scheme 9

structure analysis of the tetramethyl ether (P)-22 of mastigophorene A [(P)-1] (see Figure 3). From the relative configuration of (P)-22 in the crystal, together with the known17 S-configuration in the cyclopentyl residue, as determined above again by the CD calculations, the absolute axial configuration could be established directly to be P, which confirms the configurational assignment by Fukuyama et al. based on the exciton chirality rule.2 The synthetic approach to the mastigophorenes A [(P)-1] and B [(M)-1] presented here is most flexible and efficient, with the exception of the coupling step to the lactone 17. Unexpectedly and in contrast to so many other related coupling reactions, which gave excellent yields,6 even against highest steric hindrance (e.g., with a tert-butyl group next to the axis),18 no further optimization was possible by variation of the catalyst, the solvent, or other reaction conditions. The search for a reason for this failure led us to compare the substrate structures of successful coupling reactions in the literature6,23,24 with those of 17 and others that had previously given insufficient coupling yields.4,25,26 Strikingly, all the unsuccessful coupling substrates had a methoxy group ortho to the halogen. Indeed, previous investigations by Dyker et al.27 had shown that Pd0-catalyzed reactions of substrates such as 23 (Scheme 9) give “trimeric” products of type 24, apparently by CH activation of the O-methyl group of 25, leading to the five-membered palladacycle 26, followed by the addition of a second substrate molecule to give 27. Ring closure and involvement of a third molecule of 23 ultimately produce the observed product 24. These results suggested that, in our reaction, intermediates such as 28 and 29 (Figure 4) might play a role and be responsible for the poor coupling yield. However, there was no evidence for such intermediates and consequent byproducts. Therefore, we performed the reaction with the simplified carboxylic ester 30, i.e., in the absence of an intramolecular coupling partner (Scheme 10), thus increasing the chance of detecting possibly relevant CH activation products in our system. Indeed, with Pd(OAc)2 as the catalyst, 30 gave the cyclic biaryl ether 31 in 10% yield, consistent with initial oxidative addition (23) Bringmann, G., Saeb, W.; Ru¨benacker, M. Tetrahedron 1999, 55, 423. (24) Bringmann, G.; Ochse, M.; Go¨tz, R. J. Org. Chem. 2000, 65, 2069. (25) Bringmann, G.; Walter, R.; Weirich, R. Angew. Chem., Int. Ed. Engl. 1990, 29, 977. (26) Rao, A. V. R.; Chakraborty, T. K.; Joshi, S. P. Tetrahedron Lett. 1992, 33, 4045. (27) (a) Dyker, G. Chem. Ber./Recl. 1997, 130, 1567. (b) Dyker, G. Angew. Chem., Int. Ed. Engl. 1999, 38, 1698. (c) Dyker, G. Chem. Ber. 1994, 127, 7, 739.

9132 J. Am. Chem. Soc., Vol. 122, No. 38, 2000

Bringmann et al. Scheme 11

Figure 4. Possible intermediates in the Pd-assisted intramolecular coupling of 18.

Scheme 10

of Pd0 and subsequent cyclometalation by CH activation at the neighboring methoxy group to give 32. Addition of a second molecule of 30 and reductive elimination, with formation of the biaryl axis, apparently then lead to the intermediate 33, whose ring closure with decarboxylation would produce the ether bridge in 31. The constitution of 31 was proved, inter alia, by a series of nuclear Overhauser enhancement spectroscopy (NOESY) interactions. In addition, dynamic nuclear magnetic resonance (DNMR) investigations showed the expected splitting of the signals of the diastereomeric protons in the ether bridge (below ca. -35 °C), and thereby confirmed the chiral (though, at room temperature, configurationally unstable) biaryl structure of 31.28 This experiment demonstrates that in the bromobenzoic acid 18, it is likely that, after oxidative addition, a CH-activating interaction of Pd with the adjacent ortho-methoxy group will indeed play an important role in diminishing the yield of the desired coupling product. As a consequence of these results, a methoxy group (or any other protecting group that might undergo such CH activation) should be avoided in the immediate neighborhood of the halogen substituent. This condition had been satisfied for our successful substrates, e.g., for 1-bromonaphthalene-2-carboxylic acids and for many precursors of numerous biaryl alkaloid syntheses.6,24 Therefore, a better coupling substrate than 18 should, for example, be the modified bromo ester 36 (Scheme 11), in which the ortho-methoxy group of 18 is replaced by a diphenylmethylene acetal group for the simultaneous protection of both phenolic hydroxy functions, so that the critical ortho-OCH moieties are absent; moreover, according to experience in the literature,29,30 its deprotection (28) Line-shape analyses of the 1H NMR spectra at different low temperatures gave the corresponding rate constants k and the activation parameters ∆Hq ) 43.5 ( 1.7 kJ mol-1, ∆Sq ) -20.8 ( 7.2 J K-1 mol -1, and ∆Gq298 ) 49.7 kJ mol-1. (29) Feldman, K. S.; Ensel, S. M. J. Am. Chem. Soc. 1994, 116, 3357.

should be achieved very easily, by hydrogenolysis, even at the level of the sensitive lactone 37. The required acetal-protected bromo acid 35 31 was synthesized from herbertenediol (3), by reaction with Ph2CCl2, bromination, and cobalt-catalyzed oxidation with O2. Subsequent DCC activation and reaction with the phenol 19 gave the desired bromo ester 36. Its coupling was now highly efficient under the standard reaction conditions and our considerations presented above were confirmed. The lactone was obtained with an excellent yield of 87% and no side products or even hints of decomposition were observed. Unexpectedly however, and in contrast to results from the literature,29,30,32 cleavage of the diphenylmethylene acetal group was more difficult than anticipated. Thus, upon attempted hydrogenolysis with H2 in THF with Pd/C as the catalyst29 or by acid-catalyzed hydrolysis,32 no reaction was observed. Only after further optimization did deprotection succeed hydrogenolytically using THF/concentrated H2SO4 (10:1) as the solvent, followed by renewed protection with MeI, to give the lactone 17 in low yield. Although the overall yield of the total synthesis could not thus be improved, this does not diminish the value of our finding that avoiding an OCH unit next to the palladation site will increase the coupling yields dramatically. This important result should find further applications in other syntheses. Conclusions The work presented describes a new synthetic pathway to highly enantiopure herbertenediol (3) and its rational conversion to mastigophorene A [(P)-1] or, optionally, mastigophorene B [(M)-1] in high diastereoselectivities. Although for the construc(30) (a) Lu¨tolf, W. L.; Prewo, R.; Jost, H.; Ru¨edi, P.; Eugster, C. H. HelV. Chim. Acta 1985, 860. (b) Kawase, M.; Sinhababu, A. K.; McGhee, E. M.; Milby, T.; Borchardt, R. T. J. Med. Chem. 1990, 33, 2204. (31) This reaction sequence was first optimized with a simplified model system, bearing a tert-butyl group instead of the chiral cyclopentyl residue; for the characterization of these compounds see the Supporting Information. (32) (a) Feldman, K. S.; Quideau, S.; Appel, H. M. J. Org. Chem. 1996, 61, 6656. (b) Horie, T.; Tsukayama, M.; Kourai, H.; Nakayama, Y.; Nakayama, M. Chem. Pharm. Bull. 1986, 34, 30. (c) Bachi, M. D.; Klein, J. J. Chem. Soc., Perkin Trans. 1 1983, 1925.

Kinetic Resolution of Lactones tion and atropo-diastereodivergent oxazaborolidine-assisted cleavage of the biaryl lactone 17, the lactone concept was applied for establishing axial chirality, this methodology was extended here for the first time to the synthesis of enantiomerically pure centrochiral compounds, by highly selective (krel > 300) kinetic resolution of the aromatic-aliphatic lactone 10. A further improvement of the lactone methodology described here was to recognize and avoid possible side reactions arising from a 2-bromo-3-methoxy benzoate precursor; the methoxy group was replaced by a unit without the critical OCH entity, here a benzylidene acetal, by which the coupling yield was increased from 39 to 87%. Experimental Section Kinetic Resolution of (rac)-10. From a solution of (S)-11 (9.24 mL, 9.24 mmol, 1 M in toluene) the solvent was removed in vacuo and the residue was dissolved under argon in dry THF (50 mL). Then BH3‚ THF (15.4 mL, 15.4 mmol, 1 M in THF) was added and the solution was stirred for 30 min. This solution was added slowly to a cooled (-78 °C) solution of (rac)-10 (4.00 g, 15.4 mmol) in dry THF (200 mL) under argon. This mixture was stirred for 20 h at -78 °C, then acidified with 2 N HCl and the solvent was removed after warming to room temperature. The white precipitate was filtered off and washed with diethyl ether. The filtrate was extracted with diethyl ether, the extract was dried (MgSO4), and the solvent was evaporated. Column chromatography on silica gel (petroleum ether/diethyl ether 2:1) afforded (3aR,9bR)-cyclopentyl[d]-3a,9bH-6-methoxy-cis-(3a,9b),8-trimethylcoumarin (R,R-10) (1.91 g, 48%, ee > 99.9%, HPLC) and (S,S)12 (1.90 g) as a diastereomeric mixture. Reduction of a small sample of (S,S)-12 with LAH gave (S,S)-14 with an ee ) 94% (HPLC). (R,R)10: mp 108 °C; [R]23D ) -23.8 (c 0.73, CHCl3); CD (ethanol): λmax (∆) 203 (+42), 220 (-34), 251 (+4), 287 (+6). Atroposelective Ring Opening of 17 with (S)-11‚BH3. From 500 µL (500 µmol) of a 1 M solution of (S)-11 in toluene, the solvent was removed in vacuo and the residue was dissolved in 2 mL dry THF. After addition of 400 µL (400 µmol) of BH3 (1 M solution in THF) and 30 min stirring, this solution was added to a solution of the lactone 17 (62.0 mg, 119 µmol) in 3 mL dry THF at 0 °C. After 4 days of

J. Am. Chem. Soc., Vol. 122, No. 38, 2000 9133 stirring, the reaction was quenched with 2 N HCl, the solvent was removed in vacuo, and the remaining aqueous layer was extracted with diethyl ether. HPLC analysis gave a dr of 97:3. Purification of the dried (MgSO4) extract by column chromatography on silica gel (petroleum ether/diethyl ether 10:1) gave 34.2 mg (55%) of (P,1′′S,1′′′S)-2-hydroxy6′-hydroxymethyl-3,2′,3′-trimethoxy-6-methyl-4,4′-di-(1,2,2-trimethylcyclopentyl)biphenyl [(P)-21] (dr > 99.9:0.1) and (as a late fraction from column chromatography) 4.00 mg (6%) of a mixture of (P)- and (M)-21 (dr ) 89:11). Improved Coupling Reaction to (1′S,1′′S)-4-Methoxy-3,8-di(1,2,2-trimethylcyclopentyl)-1-methyl-9,10-(diphenylmethylidenedioxy)-dibenzo[b,d]pyran-6-one (37). In 50 mL of freshly distilled (over CaH) dimethylacetamide (DMA) 36 (518 mg, 703 µmol), Pd(PPh3)2Cl2 (49.3 mg, 70.3 µmol), NaOAc (115 mg, 1.41 mmol), and PPh3 (14.2 mg, 70.3 µmol) were dissolved under N2. After heating (130 °C), the reaction mixture was stirred for 16 h, the solvent was removed under reduced pressure, and the residue obtained was purified by column chromatography on silica gel (petroleum ether/diethyl ether 10:1). Crystallization from dichloromethane/petroleum ether gave 402 mg (87%) of 37 as colorless crystals.

Acknowledgment. We thank Dr. M. Heubes for carrying out the DNMR investigations. For making available samples of natural (P)-1 and (M)-1, we thank Professor L. J. Harrison, National University of Singapore. This work was supported by the Deutsche Forschungsgemeinschaft (SFB 347 “Selektive Reaktionen Metall-aktivierter Moleku¨le”) and by the Fonds der Chemischen Industrie. Supporting Information Available: Detailed experimental procedures and characterization data for all new compounds (including unnumberd intermediates), X-ray crystallographical data of (rac)-10, (P)-17, (P)-22, NOESY correlations of 31, and characterization of the model compounds of 34-37 (PDF). This material is available free of charge via the Internet at http://pubs.acs.org. JA001455R